The Improvement And Application Of Hybrid Frog Leaping Algorithm Based On Gradient Information

In 2003,Eusuff and Lansey proposed a hybrid shuffled frog leaping algorithm(SFLA)for the first time in order to solve the problem of minimizing the size of pipelines when the pipeline network was expanded.The SFLA algorithm is a new heuristic algorithm based on group intelligence,which is the result of modeling and simulation of Eusuff and Lansey inspired by the frog foraging behavior.The SFLA algorithm combines the advantages of the genetics-based meme algorithms and social behavior-based particle swarm optimization.The SFLA algorithm only has ten years development time since proposed,there are many aspects and areas to be studied,so the research of SFLA algorithm has a very broad prospects.Three improved hybrid shuffled frog leaping algorithms have been proposed to solve the unconstrained optimization problems,which is considered from three perspectives: poor individuals,elite individuals and the combination of poor and superior individuals.The first improved algorithm updates for the poor individuals in the group directly using the conjugate gradient method,which will get a better solution inevitably.The second improved algorithm updates with the conjugate gradient method,which will use the advantages of the elite individuals and it will enhance the elite individual's ability to guide other individuals.The third improved algorithm combines the first and the second two improvements,which update to the poor and the elite individuals at the same time.The three improved algorithms we proposed introduce the conjugate gradient method into the SFLA algorithm,which combines the strongly global searing ability of SFLA algorithm with the fast local searching ability of conjugate gradient method.We do the experiment using the international common standard problems,numerical experiments show that,comparing to the basic SFLA algorithm and other intelligent algorithms,the three improved algorithms have a higher accuracy.A hybrid algorithm for cooperative Lagrange multipliers has been presented for solving the constrained optimization problems,which is called LA_SFLA algorithm.The LA_SFLA algorithm transforms the constrained optimization problems into a series of unconstrained optimization problems,which is start from the Lagrange function of the problems based on the Augmented Lagrange multiplier method.It will solve the boundary constrained optimization problems using the improved algorithm we proposed before.We do the experiment using the common constrained problems,which will prove the LA_SFLA algorithm have a higher accuracy.Then,we apply the LA_SFLA algorithm to three engineering constraint optimization problems,numerical experiments show the effectiveness of the LA_SFLA algorithm,which will also prove the LA_SFLA algorithm has a higher application.